Precision and accuracy of subjective time estimation in different ...

Cognitive Brain Research, 1 (1993) 87-93

87

Elsevier Science Publishers B.V.

BRESC 30005

Precision and accuracy of subjective time estimation in different memory disorders Paolo Nichelli a*b,Annalena Venneri b, Mariangela Molinari b, Federica Tavani b and Jordan Grafman a a Cognitive Neuroscience Section, Medical Neurology Branch, NINDS, NIH, Bethesda, MD 20892 (USA) and b Clinica Neurologica, Universitd degli Studi di Modena, Modena (Italy)

(Accepted 27 October 1992)

Key words: Time perception; Neuropsychology; Amnesia; Frontal lobe; Alzheimer’s dementia

The aim of this study was to evaluate how different memory disorders affect subjective time durations. For this purpose we studied prospective time estimations in 4 amnesic (A) and in 15 Alzheimer’s disease (AD) patients, and compared their performance with that of 5 matched young normal controls (YC) and 15 elderly subjects (EC). For the short-time durations we asked the subject to repeatedly reproduce a standard interval of 1 s. To test how subjects evaluated longer time durations, we choose a verbal estimation procedure. The subjects’ task was to read either 5, 10, 20, or 40 digits appearing one at a time, while concurrently keeping the rhythm of 1 key press per second. At the end of each sequence, subjects had to judge the elapsed time from the beginning of the trial. Results showed that amnesics can correctly reproduce l-s intervals. However, their accuracy of verbal estimates of longer durations was severely impaired. AD patients showed increased variability on repeated reproduction of l-s intervals and were both inaccurate and imprecise in their verbal estimate of longer durations. Using the framework of the Scalar Timing Model, we conclude that amnesic patients exhibit a deficit in encoding and storing the current time for intervals that exceed their short-term memory range, while AD patients show a pattern of deficit that is explained by a more widespread involvement of both the clock, the memory, and the decisional mechanisms.

INTRODUCTION

A number of studies have shown that whenever humans have to evaluate time durations in the range of minutes or hours they depend on memory mechanisms. According to 0rnstein24 remembered durations depend on the amount of information (the ‘storage size’) stored during the relevant interval. He assumed that if more stimuli occur or if stimuli are coded in a more complex way, the experience of duration lengthens. On the contrary, the more efficient the encoding, the smaller the storage size, and the shorter the subjective duration. Alternatively, the judged duration of an interval may be a function of an encoding balance between two processors that work in parallel: a temporal and a non-temporal information processor. Thomas and coworkers29,30 restricted the range of applicability of this

model to durations less than 100 ms, but Michon” gave an example of the way the same model might be used to explain phenomena involving longer durations. In this framework, as in those previously mentioned, information retrieved from a time period is a major component of subjective time estimation of the duration of the period. Internal-clock models similar to the one proposed by Treisman 32 have been adopted by many studies aimed at understanding time-processing related issues in animals4*26. The Scalar Timing Theory is probably the most influential among them (see ref. 8 for review), and has proved to be extraordinarily useful in interpreting the effects of selective brain lesions in rats on timing 14-16,21-23.According to this model, the mean subjective time is a linear function of real time with the scalar property of variance (i.e. with a linear relationship between standard deviation of subjective time and

Correspondence: P. Nichelli, Cognitive Neuroscience Section, MNB, NINDS, NIH, Bld 10, Rm 58209 Bethesda, MD 20892, USA. Fax: (1) (301) 480-2909.

88

CLOCK

MEMORY PROCESSES

PROCESSES

Y

Fig. 1. Model of time duration processing according to the Scalar Timing Theory. The figure has been modified from Allan and Gibbon’

mean subjective time). The model also assumes that duration judgements are performed by a modular information processing system composed of clock, memory, and decisional mechanisms. Temporal judgements would be based on matching the number of accumulated pulses with values stored in reference memory. Recent studies have extended the Scalar Timing Theory to normal human subjects’,33,34.The aim of our study was to investigate duration judgements in patients with different memory disorders, taking advantage of the same theoretical framework. Indeed, we were able to find very few studies that have assessed time estimation in these kinds of patients. Richards25 examined the time reproduction of intervals ranging from 1 to 300 s by H.M., a well-known patient who became severely amnesic after a bilateral removal of He found that H.M.‘s the medial temporal cortex 18719*27. time reproductions were normal for intervals less than 20 s, but were indicative of a severe underestimation of the time passing for longer intervals. Williams, Medexamined B.W., a 55year-old wedeff, and Haban woman, who became amnesic after the removal of a dermoid cyst of the third ventricle. Results showed a severe underestimation of the time passing on both a time reproduction and a verbal estimation task. With a slightly different technique, Kinsbourne and Hicks12 examined 12 alcoholic Korsakoff syndrome patients, 12 abstinent alcoholics and 12 normal controls. The task required the subjects to crank a roll on a cash register on which a series of digits were typed. A slit showed only one digit at a time and subjects were asked to crank the roll in order to display each succes-

sive digit at a subjective one per second rate. While doing this, they had to read the number aloud. They continued for a variable number of times before being stopped by the examiner. Then they were asked to estimate the duration of the trial. The rate of digit reading multiplied by the number of digits read provided an estimate of the subjects’ judgement of the ‘present time’ (i.e. the temporal information present in immediate experience), while verbal estimates provided their judgement of the ‘past time’ (i.e. the temporal information that is out of the focus of attention and need to be recollected). The results showed that Korsakoff’s amnesics consistently underestimated the ‘past time’ for intervals longer than 30 s, but they performed as normals in judging the ‘present time’. Thus, despite sustained attention to the task and retained knowledge of how long a second is, they underestimated durations that exceeded their short-term memory. Unfortunately, Kinsbourne and Hicks12 did not give an account of the performance of each amnesic subject, making it impossible to determine if all of them underestimated the past time, or whether some subjects showed normal estimation or even overestimation of time. None of the studies on time estimation have tried to disentangle systematic from random errors, in order to make a distinction between accuracy and precision of performance. ‘Accuracy’ refers to the extent to which measurements resemble ‘real’ values of a physical quantity, whereas the term ‘precision’ is used to indicate the closeness with which measurements agree with one another, i.e. they are consistent with any specific bias (ref. 31, p. 14). As an example, if durations are measured with a clock that is running faster than the conventional clock, intervals will be inaccurate, but precise. On the contrary, if durations are measured with different clocks, each running at a different rate, results will be both inaccurate and imprecise. The aim of this study was to compare accuracy and precision of subjective time durations in patients with memory disorders of different origin. We examined patients with selective amnesic disorders and patients with Alzheimer’s disease. Later, we discuss how closely human neuropsychological data agree with results from animal studies and how the different parts of the brain may contribute to duration processing. MATERIALS AND METHODS Subjects Amnesic patients

We tested 4 amnesic patients. Table I shows their age and their educational level. Details of their clinical history is given below.

89 TABLE I Description of patient and control groups Patient group

n

Age

Alzheimer’s disease Elderly controls Amnesics: F.B. G.Z. A.B. B.B. Young controls

15 15 4

66 65

6 6

41 43 23 59 46

8 15 13 13 13

5

Education

Case 1. We examined F.B. 1 year after a subarachnoid hemorrhage from a ruptured anterior communicating artery aneurysm. Before the onset of this illness, F.B. had worked as a clerk. After clipping the aneurysm, F.B. has been severely confused and disoriented for about 2 months. Then he recovered and wanted to go back to work. However, he still complained about memory problems, especially those involving visuo-spatial abilities (e.g. he found it difficult to find his way and to learn new routes). A CT scan did not show any evidence of parenchymal damage, but angiography disclosed a right frontal artero-venous malformation supplied by branches of the anterior cerebra1 artery, of the middle cerebral artery, and of the lenticulo-striate arteries. On neurological examination crania1 nerves, strength, coordination, and body sensations were normal. His Full Scale IQ on the Wechsler Adult Intelligence Scale (WAIS) was 111 (verbal IQ = 117, performance IQ = 104). His MQ on the Wechsler Memory Scale (WMS) was 98. On the Wisconsin Card Sorting Test (WCST), he quickly and correctly identified all the categories and made only four perseverative errors. A standardized neuropsychological evaluation (see Table II) did not show any language difficulty or any verbal memory problem. Spatial short-term memory (as measured with the Corsi Block Tapping Test) also was normal, but spatial long-term memory was severely impaired, as shown by the fact that he made 156 errors and did not reach the learning criterion

TABLE II Adjusted (and equivalent) scores of amnesic patients at a standardized neuropsyhological test battery * F.B.

Bisillabic Word Span Spatial Span (Corsi) Babcock’s Test Paired Associate Learning O Supraspan Spatial Learning Phonemic Fluency Semantic Fluency Visual Search Raven PM 38 l Token Test

G.Z.

A.B.

B.B.

4.5 (4) 4 (2) 11.2 (2)

4.75 (4) 3.5 (0) 1 (0)

3.25 (1) 4.25 (3) 4.5 (0)

5.5 (4) 4.75 (4) 1.25 (0)

10.7 (3)

6

6

-

24.5 34 44.25 17.5

(2) (2) (3) (1) -

0 13 30 44.25 35.25 29.5

(0) (0) (0) (2) (3) (4) (2)

4.83 15 20 27.25 32.5 33.5

(0) (0) (0) (0) (0) (4) (4)

4.11 IO 15 53 28.75 30.25

(0) (0) (0) (4) (0) (2)

* Adjusted (an equivalent) scores are based on normative values obtained by the Italian Cooperative Study on the Neuropsychology of Aging**. According to these norms, an equivalent score of 0 corresponds to adjusted scores lower than the 5% one-sided non-parametric tolerance limit (with 95% confidence) of the normal population, an equivalent score of 4 corresponds to the performances equal or better than the median value, while the values in between are obtained dividing in three equal parts the remaining normals’ distribution3. o The normative values for the Paired Associate Learning Test were computed with the same methods. Their values are reported by Novelli et al.“. l The patient A.B was examined with Raven PM47.

after 30 trials on the Maze Test (the cut-off scores for normals are 37.5 errors and 13.41 trials to criterion6). Case 2. G.Z. attempted suicide by inhaling the exhaust emissions from his car on April 1991. We examined him 3 months later because of persistent memory deficits. On neurological examination, his cranial nerves, strength, coordination, and body sensations were normal. The EEG showed a slow dysrhythmia in the left temporal area. A CT scan was normal. His Full Scale IQ on the WAIS was 106 (verbal IQ = 103, performance IQ = 110). His MQ on the WMS was 74. On the WCST he correctly identified all the 6 categories and made only 4 non-perseverative errors. Neuropsychological evaluation (see Table II) revealed a severe anterograde amnesia for both spatial and verbal material. A defective spatial span on the Corsi Block Tapping Test and diminished fluency in associating words to a phonemic cue were also present. The Maze Test was learned only after 21 trials and with 121 errors, confirming that the anterograde amnesia extended to spatial material. Other cognitive abilities, such as language and abstract reasoning, were essentially unimpaired. Case 3. A.B. suffered from two severe closed head injuries. Both were subsequent to car accidents. The first happened 5 years ago. The second one occurred 2 years ago. After both traumas he was comatose for 20 days and subsequently began a slow progressive recovery. At the time of his neuropsychological evaluation, A.B. still complained about memory deficits, urinary urgency, and minor walking difficulties. Neurological examination showed spastic hypertonia and hyperreflexia of the four limbs with a bilateral Babinski response. The CT scan only showed a small deep hypodensity in the right fronto-parietal region. The EEG showed a slow dysrhythmia in the right frontal area. His Full Scale IQ on the WAIS was 99 (verbal IQ = 112, performance IQ = 84). His MQ on the WMS was 86. Neuropsychological evaluation (see Table II) showed anterograde amnesia both for verbal and spatial material. Language was unimpaired, but there was reduced verbal fluency with both phonemic and semantic cues. Abstract reasoning (Raven test) also was unimpaired. On the WCST, he only identified 2 categories while committing 27 perseverative errors. Case 4. Seven months before his evaluation, B.B. had suffered subarachnoid bleeding from an anterior communicating artery aneurysm. After clipping the aneurysm, the patient appeared confused and disoriented. He could not recall the year, date, or season. The lack of memories often was replaced by florid confabulation. On neurological examination, cranial nerves, strength, coordination, and body sensations were normal. The CT scan showed a bilateral hypodensity of the basal forebrain. His total IQ at the WAIS was 104 (verbal IQ = 101, performance IQ = 108). His MQ at the WMS was 84. Neuropsychological evaluation (see Table II) showed a severe anterograde memory deficit, with defective fluency and abstract reasoning (Raven test), but without any major impairment of language, attention and short-term memory. On the WCST he correctly identified only 2 categories and committed 27 perseverative errors. Patients with AIzheimerS disease

We tested a group of 15 patients with a clinical diagnosis of mild to moderate probable Alzheimer’s disease (see Table I). All of them had received a standard neurological examination, an EEG and a CT scan. The diagnosis of Alzheimer’s disease (AD) was based on the clinical criteria developed by the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association”. Only patients that met these criteria were included in this study. The patients were also staged using the Clinical Dementia Rating Scale”. According to this scale, we classified 5 patients as mildly demented, while 7 patients were classified as moderatly demented. The remaining 4 patients fell between these two categories. All the patients scored 2 10 out of a possible 30 points on the Mini-Mental State Examination 7. We administered to all the patients a standardized neuropsychological test battery including Visual Search, Paired-Associate Learning, Raven Progressive Matrices, Verbal and Visuo-spatial span tasks. Normative values for these tasks were obtained from Spinnler and Tognoniz8 and by Novelli et alzO. Table III summarizes the performance of the Alzheimer patients on these tasks. Further neuropsychological testing was administered on a case by case basis. These

TABLE

III

Adjusted scores of Alzheimer disease patients at a Standardized Neuropsyhological Test Battery * P-Ass. L. (I)

Raven PM Average S.D.

15.13 3.77

Range

s-20

Controls

2 14.75

Word Span

7.39 2.50

3.40 0.65

2.5-13

2.5-13

2 7.5

Visual Search

Spatial Span

2 2.75

2.89

28.75

1.01 l-4.2.5

11.70 2.25-48.75

> 3.50

z 30

* Scores

are adjusted for age, sex, and educational level based on the normative values obtained by the Italian Cooperative Study on the Neuropsychology of Aging%. (1) The normative values for the Paired Associate Learning Test were computed with the same methods. Their values are reported in by Novelli et al.*‘. The cut-off scores of control subjects are based on the 5% one-sided non-parametric tolerance limit (with 95% confidence) of the normal pupulation.

patients were studied peared (range: 0.7-5).

about

2 years

after

their

first symptoms

ap-

Control subjects The amnesics were matched with respect to age and educational level with 5 controls, who averaged 46.8 years of age (SD. = 5.07) and 12.8 years of education (S.D. = 3.1). The control group for the AD patients consisted of 15 elderly individuals who averaged 65 years of age (S.D. = 5.6) and 5.7 years of education (S.D. = 5.6). Neither age (tZ8= 0.565, n.s.), nor education (tz8= 0.345, n.s.1 comparisons revealed between-group differences. Tests Two different tests (Repeated Time Reproduction and Verbal Time Estimation) were prepared and implemented on a MacPlus computer using MacLab’, a test authoring application that controls exposure and response times.

the contrary 10.6% and 5.3% of the intervals were eliminated from those produced by AD patients and elderly controls, respectively. Verbal Time Estimation Task The subjects’ task was to read aloud either 5, 10, 20 or 40 single digits appearing one at a time, while pressing the space bar at a subjective 1 per second rate. By pressing the space bar, the subject caused the appearance of the next digit. Reading aloud prevented the subject from counting. Three sequences of 10, 20, and 40 digits, and 2 sequences of 5 digits were administered to each subject in a pseudorandom order that never allowed more than two sequences of the same length to be presented in a row. At the end of each sequence, the computer presented on the screen a sentence asking the subject to recall the time spent on that trial. Three training trials with three sequences of different length were administered before the beginning of the experimental session to allow subjects to become familiar with the task. Ten out of the 16 AD patients failed to keep the prescribed rhythm of one space bar press per second, while reading the digits that appeared on the screen. In this case, the experimenter pressed the key and simply asked the subject to read the digits, while keeping track of the time from the beginning of the trial. For the purpose of the analysis of the Verbal Time Estimation Task, each individual response was plotted as a linear function of the actual time elapsed. ‘Accuracy’ of performance was defined as the slope of the subject’s regression (‘b’ scores), while ‘precision’ of performance (‘P’ scores) was defined as the additive inverse of the ratio between residual (unexplained) and total variance of each individual regression (i.e. 1 minus unexplained variance divided by total variance). Values of ‘P’ ranged from 0 (lowest possible precision) to 1 (best possible precision). The best possible accuracy was defined by ‘b = 1’. Whenever b < 1, the subject’s performance was indicative of underestimation of time (i.e. of a subject judging less than the actual time to have elapsed). On the contrary, b > 1 corresponded to overestimation of time. For this reason, to compare the overall accuracy of the performance of different groups of subjects, ‘b’ scores were transformed into absolute values of their differences from 1 (‘a’ scores).

RESULTS Repeated Time Reproduction Task A l-s interval was presented five times in a row. Immediately afterwards, the subject was asked to reproduce the same interval 30 times in a row by rhythmically tapping the space bar of the computer’s keyboard. Both the intervals that were presented and the ones produced by the subjects consisted of three parts: a sound signal (a ‘beep’ lasting 150 ms, followed by a visual signal (a square appearing at the center of the screen for 67 ms), followed by a blank screen. During the study phase, the blank screen lasted 783 ms, in the reproduction phase it was terminated by the subject’s pressing the space bar (see Fig. 2). Reproduced intervals were calculated from the beginning of each cycle (the start of the ‘beep’ signal). Intervals less than 500 ms or greater than 1500 ms were eliminated from calculation, since they might represent either faulty release of the keypress or defective maintenance of the experimental set. None of the amnesics produced intervals outside these limits. Only 1 interval ( = 0.6%) was cast off in the young control group. On

A

B

1

I

beep

n

blank field

150

67

703

beep 150

’

m 67

1 msec

blank field

,,,,11,,..,111111111............ variable

w

msec

Fig. 2. Repeated Time Reproduction Test. A: represents the content of the time intervals that were shown to the subjects. B: shows the content of the intervals as reproduced by the subject.

The mean and the standard deviation of the intervals produced by each subject on the Time Reproduction Test are shown in Figs. 3 and 4. Normal subjects tended to reproduce intervals shorter than a second (young controls: 769 + 35.5 ms; elderly controls: 847 f 27.6 ms). The amnesic patients F.B. and B.B.‘s reproduced intervals were within the range of the young controls’ performance. G.Z. performed slightly outside the range of young controls, but he was well within the elderly controls’ limit. A.B., who also had some minor motor problems, was the only amnesic subject performing poorly at this task. As far as AD patients are concerned, 9 out of 15 showed a standard deviation of reproduced intervals greater than elderly controls. On average, AD patients were slower than elderly controls (t,, = 2.733, P = 0.0108) and the intervals they produced showed a greater variability (t,, = 4.775, P < 0.0001). The analysis of accuracy (b) and precision (PI scores (see Figs. 5 and 6) showed defective time estimation in

91 Accuracy

1300

of Estimated Time

6 1200

I

1100. B % 5 1000. E

4

5

l G.Z.

900.

i L 5

800.

b

l 6.6.

J

4

4

. A.0.

f

-L

l F.B. 700.

0

4

4

l-

-L

-

4

#

:

l G.Z.

4

600

l

T -

b

4 l

A.B., O-

I 500 4

Amnesics

Young c

Alzheimer

Amnesics

Ekleriy C

Young C

Alzheimer

Elderly C

Fig. 3. Mean of reproduced time. Small diamonds show single subject data, bigger diamonds show data overlap.

Fig. 5. Accuracy of estimated time. Small diamonds show single subject data, bigger diamonds show data overlap.

all the amnesic patients. However, while accuracy was affected in all four amnesics, precision was essentially unimpaired in 2 of them (G.Z. and A.B.). Three amnesic patients underestimated the time elapsed (i.e. they judged it less than the actual time), while the remaining one (B.B.) overestimated it. Note that the patient B.B. also had clear-cut neuroradiological and neuropsychological evidence of frontal lobe involvement. Also note that the amnesic patient who showed the most severe underestimation of the time passing

(F.B.) had a memory disorder that was limited to visuo-spatial tasks. AD patients, as compared with elderly controls, were greatly impaired in precision of verbal time estimates 0,s = 4.545, P < 0.0001). As far as their overall accuracy was concerned Ca’ scores), AD patients performed similarly to elderly controls (t,, = 1.588, P = 0.1236). Severity of dementia, as measured by the Mini Mental State was significantly correlated to both precision (r2s = 0.746, P < 0.0001) and to accuracy (r2s = 0.487, P = 0.0064) of verbal time estimation. Furthermore, precision scores also were found to be significantly

275, 250.

g ‘= t P 6

Precision of Eatlmated Time

225.

a

g

E

4

4

200. 175.

t

+ A.B.

4

t

150.

S ‘k! B 125. : P

l G.Z.

4

*

$

4 -7

4

l

4

B.B.

4

4

A.B.

4

4

l F.B. 100. 75.

-T

-

i

4 G.Z. 50.

l B.B.

S-Y

t

f

l F.B.

l ~

Amnesics

Young c

Alzheimer

Elderly C

Fig. 4. Standard deviation of reproduced time. Small diamonds show single subject data, bigger diamonds show data overlap.

4

Amnesics

Young C

Alzheimer

Elderly C

Fig. 6. Precision of estimated time. Small diamonds show single subject data, bigger diamonds show data overlap.

92 related to several neuropsychological measures, including the Raven PM 38 (r2s = 0.679, P < 0.00011, Visual Search (r2s = 0.622, P < 0.0002>, Paired Associate Learning (ra = 0.618, P < 0.0003>, and Corsi’s Block Tapping (r2s = 0.621, P < 0.0003) tests.

DISCUSSION The results of this study indicate that memory disorders per se are strongly related to defective accuracy in time estimation, although their mere presence neither predicts any specific bias toward under- or overestimation nor it necessarily affects the precision of the estimates. Studies on animals have shown that both the frontal and the hippocampal systems are involved in temporal processing, but in complementary ways. In a typical experiment, animals are trained to expect reinforcement after a fixed interval from a signal. Response rate is plotted as a function of the time after the signal. After training, animals receive selective brain lesions. In rats, lesions of the fimbria-fornix, a major extrinsic connection of the hippocampus, produce behavioral changes indicating that they remembered the time of reinforcement as having occurred earlier than it actually did l5 . Lesions of the frontal cortex produced an opposite effect: animals perform as if they remembered the time of reinforcement as having occurred later21. Early responses of the fimbria-fornix animals can be considered equivalent to underestimation of the time passing in humans. Scalar Timing Theory’ (see Fig. 1) interprets underestimation of time after fimbria-fornix lesion as due to an increase in activity of those mechanisms directly involved in storage of information in reference memory, while the opposite overestimation of time observed after frontal lesion is interpreted as a decrease in activity of the same mechanisms16. An alternative explanation is that early responses after fimbria-fornix lesions are due to selective interference within the temporal working memory, without affecting the remembered time of reinforcement stored in reference memory. The same explanation also can hold for humans. If memory is the cognitive mechanism by which we register the passage of time, it is conceivable that a deficit in retrieval from memory causes underestimation of time. Delayed responses of animals with lesions of the frontal cortex might be due to interference within reference memory, i.e. with mechanisms that store information about the actual time of reinforcement. If a lesion makes the stored reference interval shorter

that it actually is, while the mechanism for storing the temporal information is normal, then the animal will expect reinforcement after its real duration. In our albeit small sample of amnesic patients, the one who overestimated the time had severe signs of frontal lobe impairment. On the basis of a single case it is difficult to draw any firm conclusions, but one could argue that frontal lesions in man might produce an impairment of reference memory similar to that observed in animals with the same lesions. The co-occurrence of an impairment of the current time (due to hippocampal damage), and the balance between defective reference memory and defective current time, could determine the fact that accuracy of estimation would be biased toward either over- or underestimation. For instance, underestimation of time should occur in amnesic patients with evidence of frontal lobe damage (i.e. Korsakoff’s amnesics or patients amnesics following ischemic damage from anterior communicating artery aneurysms) if both the working and the reference memory are damaged, but the former more severely than the latter. On the contrary, overestimation should occur if the reference memory deficit prevails. Within this model, defective precision can be produced both by adding a random variability to the clock (which should result in an increased variance at the Rep.eated Time Reproduction Task) or by an impairment of the decision mechanism. All but one of the amnesic patients we studied showed an impairment of both accuracy and precision. Since this impairment was independent from the ability to keep the prescribed rhythm of one bar press per second at the Repeated Time Reproduction Task, a defective decision mechanism can better account for their imprecision. AD patients typically showed both increased intratap variability in Repeated Reproduction of short intervals and a lack of precision in Verbal Estimation of long intervals. The reproduced intervals were too few to allow an analysis in terms of motor as opposed to the clock delay I1. However, none of the AD patients we tested had any motor deficit so that a lack of precision in these patients could be due to a malfunctioning clock. In rats16 lesions of the nucleus basalis magnocellularis, an arka of the basal forebrain that has significant projections to the frontal cortex and where substantial degeneration is found in patient with Alzheimer disease2*36,produce the same effect as frontal lesions, i.e. an overestimation of time intervals. However, in AD patients one might observe both under- and overestimation or no bias, but almost always in the framework of imprecision. Therefore, we would claim that in AD

93 patients, both the clock, the memory, and the decisional mechanisms might be impaired. A final observation concerns the relative role of spatial as opposed to verbal memory deficits in determining underestimation of the time passing. In our sample of amnesic patients, we noted that the one who had the most severe underestimation of the time passing had a memory disorder that involved only the visuo-spatial material. One might therefore argue that subjective duration depends more on spatial than on verbal memory. However, in order to provide a more definite answer to this question, one should also examine patients with memory deficits restricted to verbal material. Further and more extensive studies, particularly of patients undergoing temporal lobectomy, might help to better clarify this issue. Acknowledgements.

This research was partly supported by the Italian National Research Council (Progetto Finalizzato Invecchiamento, INV 924229). We thank the Neurology Department of the University of Modena (Italy) and the Medical Neurology Branch of the National Institute of Neurological Disorders and Stroke (USA) for providing the facilities that allowed us to prepare the manuscript.

REFERENCES Allan, G.A. and Gibbon, J., Human bisection at the geometric mean, Learn Motiu., 22 (1991) 39-58. Candy, J.M., Perry, R.H., Perry, E.K., et al., Pathological changes in the nucleus of Maynert in Alzheimer’s and Parkinson’s disease, .I. Neurol. Sci., 54 (1983) 277-289. Capitani, E. and Laiacona, M., Aging and psychometric diagnosis of intellectual impairment, Deu. Neuropsychol., 4 (1988) 325-330. Church, R.M., Properties of the internal clock. In J. Gibbon and L. Allan (Eds.), Annals of the New York Academy of Sciences: Timing and Time Perception, Vol. 423, The New York Academy of Sciences, New York, 1984, pp. 566-582. Costin, D., MacLab: a Macintosh system for psychology labs, Behav. Res. Methods Instrum. Comput., 20 (1988) 197-200.

De Renzi, E., Faglioni, P. and Villa, P., Topographical amnesia, J. Neural. Neurosurg. Psychiatry, 40 (1977) 498-505.

Folstein, M.F., Folstein, S.E. and McHugh, P.R., Mini-mental state. A practical method for grading the cognitive state of patients for the clinician, J. Psychiatry Res., 12 (1975) 189-198. Gibbon, J., Origins of scalar timing, Learn Moth., 22 (1991) 33-38. 9 Gibbon,

J., Church, R.M. and Meek, W.H., Scalar timing in memory. In J. Gibbon and L. Allan (Eds.1, Annals of the New York Academy of Sciences: Timing and Time Perception, Vol. 423, The New York Academy of Sciences, New York, 1984, pp. 52-77. 10 Hughes, C.P., Berg, L., Dazinger, W.L., Coben, L.A. and Martin, R.L., A new clinical scale for the staging of dementia, Br. J. Psychiatry, 140 (1982) 566-572.

11 Ivry, R.L. and Keele, SW., Timing functions of the cerebellum, J. Cogn. Neurosci., 1 (1989) 136-152. 12 Kinsbourne, M. and Hicks, R.E., The extended present: evidence from time estimation by amnesics and normals. In G. Vallar and T. Shallice (Eds.), Neuropsychological Impairments of Short-Term Memory, Cambridge University Press, Cambridge, 1990, pp. 319329.

13 McKann, G., Drachman, D., Folstein, M., Katzman, R., Price, D. and Stadlan, E.M., Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA work group under the auspicies of the department of health and human services task force on Alzheimer’s disease, Neurology, 34 (1984) 939-944. 14 Meek, W.H., Hippocampal function is required for feed-back control of an internal clock’s criterion, Behau. Neurosci., 102 (1988) 54-60. 15 Meek, W.H., Church, R.M. and Olton, D.S., Hippocampus, time, and memory, Behau. Neurosci., 98 (1984) 3-22. 16 Meek, W.H., Church, R.M., Wenk, G.L. and Olton, D.D., Nu-

cleus basalis magnocellularis and medial septal area lesions differently impair temporal memory, J. Neurosci., 7 (1987) 35053511. 17 Michon, J.A., The compleat time experiencer. In: J.A. Michon and J.L. Jackson (Eds.), Time, Mind, and Behavior, Springer,

Berlin, 1985, pp. 20-52. 18 Milner, B., Les troubles de la memoire accompagnant des lesion hippocampiques bilaterales. In P. Passuant (Ed.), Physiologie de l’tiippocampe, Centre National de la Recherche Scientifique, Paris, 1962, pp. 257-272. 19 Milner, B., Corkin, S. and Teuber, H.-L., Further analysis of the hippocampal amnesic syndrome: 1Cyear follow-up study of H.M, Neuropsychologia, 6 (1968) 215-234. 20 Novelli, G., Papagno, C., Capitani, E., Laiacona, M., Cappa, S.F.

and Vallar, G., Tre test clinici di memoria verbale a lunge termine: taratura su soggetti normali, Arch. Psicol. Neural. Psichintr., 47 (1986) 477-506. 21 Olton, D.S., Frontal cortex, timing and memory, Neuropsychologiu, 27 (1989) 121-130. 0* Olton, D.S., Meek, W.H. and Church, R.M., Separation of hip‘L

pocampal and amigdaloid involvement in temporal memory dysfunction, Bruin Res., 404 (1987) 180-188. 23 Olton, D.S., Wenk, G.L., Church, R.M. and Meek, W.H., Attention and the frontal cortex as examined by simultaneous temporal processing, Neuropsychologia, 26 (1988) 307-318. 24 Ornstein, R.E., On the Experience of Time, Penguin, Harmondsworth, UK, 1969. 25 Richards, W., Time reproductions by H.M., Acta Psychol., 37 (1973) 279-282.

26 Roitblat, H.L., Introduction to Comparative Cognition, Freeman, New York, 1987, 377 pp. 27 Scoville, W.B. and Milner, B., Loss of recent memory after bilateral hippocampal lesions, J. Neurol. Neurosurg. Psychiatry, 20 (1964) 11-21.

28 Spinnler, H. and Tognoni, G., Standardizzazione e taratura italiana di test neuropsicologici [Normative values of some neuropsychological test for Italian population], Ital. J. Neurol. Sci., Suppl. 8 (1987) 120. 29 Thomas, E.A.C. and Brown Jr., I., Time perception and the filled-duration illusion, Percept. Psychophys., 16 (1974) 449-458. 30 Thomas, E.A.C. and Weaver, W.B., Cognitive processing and time perception, Percept Psychophys., 17 (1975) 363-367. 31 Topping, J., Errors of Observation and Their Treatment, Chapman and Hall, London, 1955, 119 pp. 32 Treisman, M., Temporal discrimination and the indifference interval: implications for a model of the ‘internal clock’, Psychol. Monogr., 77 (1963) 1-31.

33 Wearden, J.H., Do humans possess an internal clock with scalar timing properties?, Learn Moth., 22 (1991) 59-83. 34 Wearden, J.H., Human performance on an analogue of an interval bisection task, Q.J. Exp. Psychol., 43 (1991) 59-81. 35 Whitehouse, P.J., Price, D.L., Struble, R.G., Clark, A.W. and DeLong, M.R., Alzheimer’s disease and senile dementia: loss of neurons in the basal forebrain, Science, 215 (1982) 1237-1239. 36 Williams, J.M., Medwedeff, C.H. and Haban, G., Memory disorders and subjective time estimation, J. Clin. Exp. Neuropsychol., 11 (1989) 713-723.